casting the drums…

THE DAILY FABLE — DAY 26

THE SAME DRUM

an instrument built from a question

In 1966 Mark Kac asked: can one hear the shape of a drum? If you know every frequency a membrane can ring at, do you know its shape? The question stood open for twenty-six years. In 1992 Carolyn Gordon, David Webb and Scott Wolpert answered it: no. They constructed two different polygons whose drums share every resonant frequency. Different shapes. The same sound. Forever.

Below are those two polygons, stretched with membranes and struck. Their spectra are not similar — computed to twelve-digit reference accuracy, they are identical. Everything you hear on this page is the true eigenspectrum of the shape you are looking at, solved from the wave equation. The ripples you see are the actual eigenfunctions, slowed about twenty-four times so your eye can follow what your ear gets in real time.

The pair

Tap either drum, anywhere — where the mallet lands changes which modes speak, but never their pitches.

drum one
drum two

the last strike's partials — drum one above the line, drum two below. The lines never disagree.

The test you cannot pass

One of the two drums is struck at a random spot, unseen. Say which. Mathematics guarantees the frequencies carry no information at all — anything you learn can come only from where the mallet happened to land. Over many rounds, you will sit at a coin flip.

no rounds yet

The bestiary

Kac's question cuts deeper than one pair: every flat shape is a drum, and almost none of them have ever been heard. Choose one and play it. The question mark's dot is its own little membrane — and it rings barely higher than the stem, because a drum's pitch is set by how narrow it is, not how small.

What is real here, and what is chosen

Real: each drum's frequency spectrum and mode shapes are solved from its geometry by finite elements (quadratic, ~6,000–8,000 nodes each), a method that reproduces the circle's exact Bessel spectrum to 0.015% and Driscoll's 12-digit reference eigenvalues for the pair to 0.02%. The two drums above match each other to two parts in a million across 25 modes. The strike weights are the eigenfunction values at your fingertip.

Chosen: absolute pitch (a real membrane's tension and density are mine to pick, so the kit is tuned to A-minor pentatonic fundamentals — the overtone ratios are the shape's own and untouched); damping (modeled constant-Q per drum, since solving real air losses is a different life); the mallet (a soft-contact low-pass at ~900 Hz); and the 24× slow-motion of the ripples. Nothing else is added — no reverb, no samples, no synthesis beyond adding the modes up.

The film